The present invention relates to a cell discard method for discarding of buffered fragmented frames within a communications network as described in the preamble of claim 1, and related devices as described in the preambles of claim 4 and claim 6.
Such a cell discard method is already known in the art, e.g. from the section xe2x80x9cTraffic and Congestion Control in ATM Networksxe2x80x9d, pages 302-340 in Asynchronous Transfer Mode: Solution for Broadband ISDN, third edition, from the author Martin de Prycker. Therein, a cell based traffic policing method called Usage Parameter Control (UPC), containing the steps of checking cells called determining cells to be conforming or non-conforming and discarding cells called cell discarding is described.
Since in a cell based discard method only single cells from a frame are discarded, the frame is left incomplete in the network. Since incomplete frames are useless from an application point of view, hence the presence of these frames decreases network throughput. This situation can be improved by using frame based traffic policing methods, but most of the times these have the disadvantage, that a number of cells of the entire frame which should be discarded, already have left the discarding device. Only the cells that still remain within the discarding device can be discarded. A corrupted frame is left then in the communications network. If this corrupted frame is not used anywhere in the communication network, then the presence of this frames also decreases network throughput. As a remark on this frame based policing methods it should be mentioned that there are frame based policing algorithms possible and available not having the disadvantage of leaving corrupted frames within the network. One of those methods is described in the U.S. Pat. No. 5,666,353 with title xe2x80x9cFrame based traffic policing for a digital switchxe2x80x9d. However the disadvantage of this kind of methods is that the network has to support this method end-to-end to avoid incomplete frames within the network. This situation is very unlikely in the near future because it should be agreed and supported by all network operators. Summarising the former, it can be said that applying traffic policing methods in a communication network in general, involves that the frames within the communication network may be incomplete because of a number of different possible problems already mentioned. If such incomplete frames are not used anywhere in the communication network, then the presence of these frames decreases network throughput. This can also be the situation in case of cell discard by network elements like network buffers which discard cells in case of network congestion whereby also corrupted frames are left within the network.
An object of the present invention is to provide a cell discard method of the above known type and related devices adopted to perform this method, but wherein the network throughput is improved.
By marking a frame wherefrom a cell has been deleted, discarding of this frame is facilitated. In this way, all corrupted frames that usually would be transported to its destination node in the prior art system are in the present invention discarded in an early stage, i.e. in the second discarding device, thereby decreasing network load and preventing the network from overload. To discard a marked frame completely, all leftover cells belonging to this frame should be stored in the second discarding device. If this is not the case, network load is reduced to a smaller extent by only discarding those cells which are present in the second discarding device.
However, those remaining cells of the frame, i.e. those are the cells which have left the second discarding device together with the last cell, can be discarded by another coupled second discarding device.
The first discarding device for instance can be a traffic policer. In that case, the criterion to check is a conformance criterion. Alternatively the first discarding device can be a network buffer. In that case, the criterion to check is a congestion criterion.
The corrupted frame indicator can be implemented by assigning a value to a single bit in the end of message cell which indicates the end of the corresponding frame. The meaning of the value 0 or 1 assigned to the bit is the state of the corresponding frame: corrupted or complete or vice versa. This implementation is a preferred solution because the end of message cell is often used in such communication networks whereby also spare bits are reserved for future implementations. By using spare bits of an existing cell no additional traffic is generated.